All my low noise RF preamps have a maximum input of 0 dbm and can handle the maximum output power of the nanoVNA without an attenuator. You have to know what your system budget is and choose your attenuators accordingly. I always look up the specs for the maximum gain and maximum input power of the amplifier under test prior to testing. As an example, today I tested an HP8447D RF preamp. It has a specified 25 dB gain from 100k-1.3GHz and a maximum input of 0 dBm. Knowing those facts, I chose to use at least 25 dB of attenuation in-line during testing to offset the gain of the HP8447D, and stay within the 40 db S21 dynamic range of the nanoVNA at 900 MHz. If you are testing below 300 MHz you can get away with a lot more attention in-line (S21 dynamic range is spec'd at 70 dB). I ended up using a 10 dB and 20 dB attenuator in-line. The 10 dB attenuator was connected to the CH0 output of the nanoVNA and the input of the HP447D. The HP8447D can handle the output of the nanoVNA without an attenuator, but it is good practice to use an attenuator to lower the mismatch and provide isolation. The 20 dB attenuator was connected to the output of the preamplifier and the CH1 input of the nanoVNA . As with the other RF preamps I've tested using the same procedure, the system operated linearly over the 50k-900 MHz test range and the measured gain of the HP8447D was 26 dB =/- 1 dB. This result was verified using a calibrated HP8753C VNA.
The above is my typical S21 measurement procedure for verifying the gain of a low noise RF preamp, such as one that might be used to improve the noise floor of a receiver or spectrum analyzer. When testing a block LNA more attenuation, as suggested by qrp.ddc, may be in order.
